Complication avoidance in microsurgery (neurosurgery, opthalmology, otorhinolaryngology, limb and digit reattachment) is crucial, and minimizes patient morbidity and health care costs. Current operative techniques rely on human surgeons, who have variable skill and dexterity. They also have physiological limits to their precision, tactile sensibility and stamina. Furthermore, the precise localization of brain pathology and neural structures is often difficult to achieve during surgery due to brain shifts and deformations as the operation progresses. While Intra-operative Magnetic Resonance Imaging (iMRI) has been used to monitor brain deformations, the surgeon currently has no effective way to use the iMRI data to enhance the precision and dexterity of surgery. They are compelled to rely on old techniques, and do not take advantage of these exquisite, updated images. Consequently, the quality of the surgery and outcomes is variable, and too often sub-optimal.
Surgical robots have the potential to increase the consistency and quality of neurosurgery, and when used in conjunction with the advanced diagnostic imaging capabilities of iMRI, can offer dramatic improvements. Unfortunately, there are no surgical robots that provide the surgeon with an ambidextrous and precise surgical system that uses updated iMRI patient data to achieve accurate image-guided surgery. In addition, there is no surgical robot with force sensing technology that is compatible with MRI systems.
Traditional surgery relies on the physician's surgical skills and dexterity and ability to localize structures in the body. Surgical robots have recently been developed to address the physical human issues such as fatigue and tremor in procedures. These systems were specifically developed for Minimally Invasive Surgery (MIS) or “key-hole” general surgery, orthopaedics and stereotactic neurosurgery.
The Intuitive Surgical Inc. DA VINCI® and Computer Motion ZEUS® robots are examples of MIS robots. MIS robots are not suitable for neurosurgery since they require a portal in the body and lack the required dexterity and ability to reposition to different surgical worksites. Furthermore, neither system is MR compatible nor is there any force feedback capability. One patent on this development is U.S. Pat. No. 6,394,998 of Wallace et al issued May 28th 2002.
The DA VINCI® system is archetypal for general surgical robots. It has an articulated ENDOWRIST®, at the end of its two 7 mm diameter ‘working’ arms. A more stable camera arm with two lenses (allowing stereoscopic images) is also inserted through an 8 mm portal. The end-effectors can manipulate instruments with tips as small as 2 mm. They have seven degrees of freedom (three at the wrist). The surgeon controls the robot through a console placed in the operating room, allowing control of both the external and internal surgical environments. The surgeon's interface has instrument controllers that can filter tremor and decrease the scale of motion. Foot pedals expand the surgeon's repertoire, allowing tissue coagulation and irrigation. Visual feedback is through a proprietary stereoscopic display, called Surgical Immersion™. FDA approval has been obtained for thoracoscopic, dominal and prostate procedures. Over one hundred DA VINCI® systems have been sold, and have been used to perform cholecystectomies, Nissen fundoplications, adrenalectomies, nephrectomies, mitral valve repairs, coronary artery bypass grafting and prostatectomies.
Surgical robots in orthopaedics may be classified as positioning or machining aids. ROBODOC®, used for hip replacement surgery, is an example of the latter. Again, they lack the dexterity, MR compatibility and force sensing needed for neurosurgery. The first-generation ROBODOC® was developed by IBM and the University of California Davis campus. The system was initially tested on 26 dogs in 1990. A second-generation ROBODOC® was built by Integrated Surgical Systems, and human trials conducted. In contrast, Kienzle developed a positioning device for total knee replacement (TKR). It locates the tibia and femur, and correctly positions the drill guide for the surgeon. Guide blocks are inserted into the drill holes, allowing the surgeon to accurately prepare the patient's bones for joint implantation. A similar system, named the Acrobot, has been developed by the Imperial College group and is designed for accurate machining of bone surfaces in TKR surgery. All the systems mentioned depend on preoperatively placed fiducial markers. Patents on this development are U.S. Pat. Nos. 5,695,500; 5,397,323 (both Taylor); U.S. Pat. Nos. 5,086,401 and 5,408,409 (both Glassman) issued in 1992 to 1997.
Robots designed for neurosurgical applications are generally restricted to positioning and holding instruments for simple procedures such as stereotactic biopsies.
In 1991, Drake reported the use of a PUMA 200 robot as a neurosurgical retractor in the resection of six thalamic astrocytomas. It is the same machine that was first used by Kwoh in 1985 to perform stereotactic biopsies. The robot has revolute joints and has six DOF. Individual joints are moved by DC servomotors, and their position and velocity tracked by optical encoders. The robot arm could be programmed to move into position, or manually manipulated in a passive mode. Its repeatability was measured at 0.05 mm, and error of accuracy at 2 mm. Its pneumatic gripper was used to clasp a brain retractor only. The cases were all performed with a BRW stereotactic frame in place, secured to the same rigid structure as the PUMA arm. This allows for stable transformation of stereotactic to robotic coordinates. Target coordinates were transferred to a computer work station with 3D CT images, enabling the brain retractor to be accurately placed in relation to the lesion. Progress in developing this system was limited by the inability to rapidly render updated 3D brain images in the operating room. The recent convergence of advanced computing, software and iMR imaging now allows us to initiate sophisticated neurorobotics.
A six DOF robot has also been used by Benabid from 1987 to position brain cannulae. It is attached to a stereotactic frame, and can use spatially encoded data from Xray, CT, MR imaging and angiography to plot its path. These images are also fused with digitized brain atlases to assist in surgical planning. Hundreds of stereotactic cases have been performed, including endoscopy (1-3). Similarly, URS (Universal Robotic Systems) has developed a six DOF hexapod robot called Evolution 1 for brain and spinal surgery. This system is based on a parallel actuator configuration, which provides it with high positional accuracy and large payload capacity. The positional accuracy is essential for stereotactic procedures and the high payload capacity may make Evolution 1 particularly well suited for drilling applications such as pedicle screw placements in the spine.
A simulation tool for neurosurgery, ROBO-SIM, has recently been developed. Patient imaging data is entered and the surgical target and corridor can be selected and planned. Virtual constraints are determined, creating no-go zones. The system can be connected to a robotic arm, NEUROBOT, which holds and positions an endoscope for the surgeon. NEUROBOT has four degrees of freedom if pivoted around the burr-hole. At this time, there are no published reports of it being used on patients. It is attached to a stereotactic frame, and can use spatially encoded data from Xray, CT, MR imaging and angiography to plot its path. Again, the systems have only one robotic arm and cannot emulate a human surgeon.
A dextrous robot called the Robot-Assisted Microsurgery system (RAMS), was developed by NASA's Jet Propulsion Laboratory. The mechanical subsystem is a six-DOF robot slave arm driven by tendons. This allows a large work envelope. It is designed to have 10 microns positioning accuracy. The master input device also has six tendon-driven joints. Simulated force feedback has been used, and it has potential to be used tele-robotically. RAMS is capable of being used to enhance various types of microsurgery, including opthalmology. Although RAMS has the required dexterity, it is still a single arm system lacking the ability to reposition itself over a large worksite. It is also not MR compatible and has no direct force feedback sensing capability and is not image-guided. Patents on this development are U.S. Pat. Nos. 5,784,542; 5,710,870; 6,385,509 and 6,233,504 all of Das and Ohm et al issued in 1998, 2001 and 2002.
The only MR compatible ‘robot’ is a simple experimental system developed by Chinzei and at the Brigham and Women's Hospital in Boston, USA. The robot consists of a passive instrument holder attached to Cartesian translational stages. The limited capabilities of the device caused it to fall into disuse.
The progress of clinical neurological sciences has depended on accurate cerebral localization and imaging technology. Over the past century, advances in cerebral imaging including contrast angiography, pneumoencephalography, and in more recent decades, ultrasound imaging, CT, MRI and frameless stereotactic navigation technology have revolutionized cerebral localization. Neurosurgery's dependence on imaging technology is epitomized by the recent flurry of iMR imaging systems developed to provide MR images during a neurosurgical procedure. Since 1996, multiple MR systems and related technologies have been developed, with over 3000 neurosurgical procedures performed worldwide. The systems possess magnet field strengths ranging from 0.12 to 1.5 Tesla, associated with varying degrees of intrusion into standard neurosurgical, anaesthetic and nursing procedures and protocols.